def test_parse_contig_descritpion(self):
id_string = "my_genome_name_uid123_100 my_family|my_genus|my_genus my_species"
contig_id_hash = parse_contig_description(id_string)
assert_equal("my_genome_name_uid123", contig_id_hash["genome"])
assert_equal(100, contig_id_hash["contig_id"])
assert_equal("my_family", contig_id_hash["family"])
assert_equal("my_genus", contig_id_hash["genus"])
assert_equal("my_genus my_species", contig_id_hash["species"])
def read_contigs_file(open_contigs_file, start_position=False,taxonomy_info=True):
""" Read contigs file generated by generate_contigs script"""
contigs = []
seqs = list(SeqIO.parse(open_contigs_file, "fasta"))
for seq in seqs:
if taxonomy_info:
contig_id_hash = parse_contig_description(seq.description, start_position=start_position)
contig = DNA(id=contig_id_hash["genome"], seq=str(seq.seq))
if start_position:
contig.start_position = contig_id_hash["start_position"]
contig.family = contig_id_hash["family"]
contig.genus = contig_id_hash["genus"]
contig.species = contig_id_hash["species"]
contig.contig_id = contig_id_hash["contig_id"]
else:
contig = DNA(id=seq.id,seq=str(seq.seq))
contig.contig_id = seq.id
contigs.append(contig)
return contigs